Spectral Clustering of Risk Score Trajectories Stratifies Sepsis Patients by Clinical Outcome and Interventions Received

Abstract

Sepsis is not a monolithic disease, but a loose collection of symptoms with a diverse range of outcomes. The diverse patterns of sepsis make guideline-driven treatment difficult, as guidelines are based on the needs of the "average" patient. Thus, stratification and subtyping of sepsis patients is of interests, with the ultimate goal of identifying groups of patients who respond similarly to treatment. To do this, we examine the temporal evolution of patient state using our previously-published method for computing patient risk of transition into septic shock. The application of spectral clustering to risk score trajectories reveals that these trajectories diverge into four distinct clusters in the time window following early prediction of septic shock. Patients in these clusters stratify by outcome: the highest-risk group has a 76.5% prevalence of septic shock and 43% mortality, whereas the lowest-risk group has a 10.4% prevalence of septic shock and 18% mortality. These clusters differ also in treatments received, as well as median time to septic shock onset. Data analyses reveal the existence of a rapid (30-60 min) transition in risk at the time of threshold crossing. We hypothesize that this rapid transition occurs as a result of an abrupt failure of compensatory biological systems to cope with infection, resulting in a bifurcation of low to high risk. Such a collapse in compensation, we believe, represents the true onset of septic shock. Thus, what we previously referred to as the pre-shock state represents a potential new data-driven definition of septic shock.

Individualized resuscitation strategy for septic shock formalized by finite mixture modeling and dynamic treatment regimen

“…Sepsis has been found to be consisted of several phenotypes, though specific class membership assignments are different across studies [12][13][14][15][16][17][18]. By using clinical trial data from 1696 patients, Gårdlund B and colleagues identified six classes of septic shock [19].…”

Early prediction of impending septic shock in children using age-adjusted Sepsis-3 criteria

“…We repeated our analyses of risk score trajectories for stratification of sepsis patients 15 on the SEQUIP dataset. Spectral clustering 21 of risk score trajectories in the window surrounding early prediction yielded two clusters (Figure 3).…”

“…Previously, using spectral clustering of patient risk score trajectories, we further elucidated the pre-shock state, finding that entry into the state was marked by a rapid transition from low to high risk. Prior to entry, sepsis patient physiology was indistinguishable between the low and high risk clusters of patients 15 . However, after the occurrence of this event, patient risk trajectories diverged, and stratified patients by risk of septic shock, mortality, time to septic shock onset, and treatments received.…”

“…Stratification of sepsis patients by risk score trajectories was performed as described previously 15 . Risk score trajectories for each patient were computed by applying the XGBoost model at each hour in the time window surrounding time of early prediction using the most recently observed values of each feature.…”

“…We developed a method for predicting an oncoming transition from sepsis to septic shock based on the hypothesis that there exists a physiologically distinct state of sepsis, which we term the "pre-shock state", and that entry into this state presages the impending onset of septic shock 13 . Through analyses of the temporal evolution of patient state, we previously stratified adult sepsis patients by outcome and interventions received, and discovered that entry into the pre-shock state was marked by a rapid shift in patient physiology occurring within a 30-60 minute timeframe 15 .…”